Problem Set 5

Due: June 13, 2025 at 11:59pm EST

Instructions for Completing the Problem Set:

• Please complete all the questions provided; however, note that some may not be graded. Unless
otherwise indicated, you must provide full proofs or justifications for your answers.
• You are welcome to reference any lemmas or theorems that were proven in the textbook or during
lectures. You may also reference lemmas or theorems that were explicitly stated but not proven, as
long as doing so does not trivialize the problem.

• Collaboration with your peers is allowed and encouraged! Working together can deepen your under-
standing. However, it is essential that you personally understand the solutions and are able to explain
them in your own words.
• All submissions must be written in your own words. This ensures you have internalized the
material and are strengthening your own reasoning and problem-solving abilities.

• Please refrain from using external tools such as AI. The purpose of these problem sets is not simply
to earn a grade, but to prepare you for the final exam and, more importantly, to help you truly
master the material. Think of each problem as an opportunity to practice, explore, and solidify your
understanding.
• All submissions should be uploaded through Gradescope by the specified deadline.

Remember: The goal here is learning, not just completing an assignment. Focus on understanding the
concepts deeply — the grades will naturally follow. We believe in your ability to grow and succeed!

Questions:
1. (a) Let g(x, y) : R2 → R2 by given by g(x, y) = (2ye2x , xey ). Show that g is a bijection between a
neighbourhood of (0, 1) and a neighborhood of (2, 0) and compute D(g −1 )(2, 0).

Solution: g is a c1 function with g(0, 1) = (2, 0) so we can use the IFT:

 2x
2e2x

4ye
Dg =
ey xey
 
4 2
Dg(0, 1) =
e 0
det(Dg(0, 1)) = 4 − 2e ̸= 0

1
 By the IVT, g is invertible locally around (0, 1). i.e g is a bijection between a neighbourhood
of (0, 1) and a neighborhood of (2, 0)

For the derivative: by the IVT, we have that:

D(g −1 )(2, 0) = Dg(0, 1)−1

 −1
−1 4 2
D(g )(2, 0) =
e 0
1
 
−1 0 e
D(g )(2, 0) = 1
2 − 2e

(b) Let f : R2 → R2 be given by f (x, y) = (x2 − y 2 , 2xy). Let A = {(x, y) : x > 0} ⊂ R2 . Show f is a
bijection between A and f (A) and compute D(f −1 )(0, 1). (Hint: if a = b then ∥a∥ = ∥b∥)

Solution: Note, we can’t apply IFT directly because the IFT can only tell us if f is a bijection
locally around a point. We instead show that f is injective and surjective. f : A → f (A) is
surjective because the co-domain is chosen to be the image pf of domain thus by definition,
every point in the co-domain is mapped to by some point in the domain. To show f is injective,
let f (a, b) = f (c, d) so (a2 − b2 , 2ab) = (c2 − d2 , 2cd) and in particular a2 − b2 = c2 − d2 and
ab = cd. Furthermore:

f (a, b) = f (c, d)
∥f (a, b)∥ = ∥f (c, d)∥
2 2
(a2 − b2 , 2ab) = (c2 − d2 , 2cd)
(a2 − b2 )2 + 4a2 b2 = (c2 − d2 )2 + 2c2 d2
a4 + 2a2 b2 + b4 = c4 + 2c2 d2 + d4
(a2 + b2 )2 = (c2 + d2 )2
a2 + b2 = c2 + d2

Adding the equations a2 + b2 = c2 + d2 and a2 − b2 = c2 − d2 gives us that a2 = c2 and since

x > 0 in A we have that a = c. Combining this with ab = cd gives us b = d. Therefore
f : A → f (A) is a bijection because it is injective and surjective.

To compute the derivative, Note that f ( √12 , √12 ) = (0, 1). Also f is C 1 so we can apply the
IFT:
 
2x −2y
Df =
2y 2x
" #
1 1 √2 − √22
Df ( √ , √ ) = √2 2
2 2 √2
2 2
1 1
det(Df ( √ , √ )) = 4 ̸= 0
2 2

Page 2
 which implies that f −1 is C 1 and the derivative is given by:
1 1
D(f −1 )(1, 0) = Df ( √ , √ )−1
2 2
" #−1
√2 − √2
D(f −1 )(1, 0) = √22 √2
2
2 2
"√ √ #
2 2
D(g −1 )(2, 0) = 4√ √4
− 42 4
2

2. Consider the set S:

S = {(x, y, z, w) ∈ R4 : x2 + y 2 + z 2 + w2 = 1, x + y = 1}

(a) Show that any point in S can be written locally as the graph of a C 1 function.

Solution: Consider the function F : R4 → R2 :

 2
x + y 2 + z 2 + w2 − 1

F (x, y, z, w) =
x+y−1

Then S is the level set F = 0. Also note that F is C 1 and the derivative is:
 
2x 2y 2z 2w
DF (x, y, z, w) =
1 1 0 0

This matrix fails to be rank 2 only if x = y and z = w = 0 (so the first row is a multiple of the
second). However if x = y, by the second equation x + y = 1 implies x = y = 21 in which case
the first equation is not satisfied. Therefore every point on S has DF be rank 2 and so every
point can be written locally as a graph of a C 1 function.

(b) Which point on the surface S is the farthest away from the point (0, 0, 2, 0)?
p
Solution: The distance of a point (x, y, z, w) on S to (0, 0, 2, 0) is given by x2 + (y)2 + (z − 2)2 + w2
so we can maximize the function g(x, y, z, w) = x2 + (y)2 + (z − 2)2 + w2 using lagrange multi-
pliers:

∇g = λ1 ∇F1 + λ2 ∇F2
(2x, 2y, 2(z − 2), 2w) = λ1 (2x, 2y, 2z, 2w) + λ2 (1, 1, 0, 0)

If λ1 = 0 then using 2(z − 2) = λ1 2z we have z = 2 and w = 0 and x = y = λ22 . Using the

second constraint we have x = y = 12 . However ( 12 , 12 , 2, 0) doesn’t satisfy the first constraint.
Now if λ1 = 1 then 2(z − 2) = λ1 2z isn’t satisifed for any z.
Thus we have λ1 ̸= 0, 1. In this case using 2x = λ1 2x + λ2 and 2y = λ1 2y + λ2 we note that
λ2
x = y = 2−2λ 1
. Also using the second constraint we have x = y = 21 . Also by 2w = λ1 2w we
p q
have w = 0. Finally using the first constraint we have z = ± 1 − x2 − y 2 − w2 = ± 12 . Thus

Page 3
 the furthest point is either ( 12 , 21 , √12 , 0) or ( 12 , 12 , − √12 , 0) and plugging this into the distance
function we see that the furthest point is ( 21 , 21 , − √12 , 0).

3. Find the maximum value for f (x, y, z) = x3 − 2xy + y 2 + 2z on S = {(x, y, z) ∈ R3 : 0 ≤ z ≤ 1 − x2 − y 2 }.

You must fully justify your solution.

Solution:
Existence of a maximum:

• Note that S is compact. Since f is a polynomial, it is continuous on S. Therefore f |S , and

hence f , has a global maximum on S by the extreme value theorem.

Searching the interior:

• Since f is differentiable on S o , the local extreme value theorem tells us that the local extrema
of f on S o occur where ∇f (x, y, z) = 0.
• We compute the gradient as ∇f (x, y, z) = (3x2 − 2y, −2x + 2y, 2). Therefore, ∇f (x, y, z) ̸= 0.

• So, by the local extreme value theorem, f has no local extrema on S o . Since global extrema
are also local extrema, the global maximum of f on S cannot occur in the interior of S.

Searching the boundary:

• We can decompose the boundary into three pieces: A = {(x, y, z) : z = 0, x2 + y 2 < 1},
B = {(x, y, z) : z = 1 − x2 − y 2 , x2 + y 2 < 1}, and C = {(x, y, z) : z = 0, x2 + y 2 = 1}.
• To optimize on A, define GA : {(x, y, z) ∈ R3 : x2 + y 2 < 1} → R by GA (x, y, z) = z. Then
GA is C 1 and its domain is open. Furthermore, A = G−1 A ({0}).

• By theorem 4.6.8, if the maximum of f on S occurs on A then it must satisfy the Lagrange
system. The Lagrange multiplier system for A is

3x2 − 2y = 0 − 2x + 2y = 0 2=λ z=0

• The solutions to this system 2

are x = y = 0 and x = y = 3 . Therefore, we must consider the
2 2
points (0, 0, 0) and 3 , 3 , 0 .

• To optimize on B, define GB : {(x, y, z) ∈ R3 : z > 0} → R by GB (x, y, z) = 1 − x2 − y 2 − z.

Note that GB is C 1 and its domain is open. To see that B = G−1 −1
B ({0}), note that B ⊆ GB ({0})
immediately. Furthermore, if GB (x, y, z) = 0, then z = 1 − x2 − y 2 and z > 0 so x2 + y 2 < 1
as well. Therefore (x, y, z) ∈ B. So G−1
B ({0}) ⊆ B.

• As for A, we check the Lagrange system. The Lagrange multiplier system for B is

3x2 − 2y = −2λx − 2x + 2y = −2λy 2 = −λ z = 1 − x2 − y 2 .

• Therefore, λ = −2. This gives y = −x and so 3x2 + 2x =
4x. The solutions to this are x = 0
and x = 32 . This gives us the points (0, 0, 1) and 32 , − 23 , 19 to consider.

Page 4
 • Lastly, to optimize over C we define GC : R3 → R by GC (x, y, z) = (x2 + y 2 − 1, z). Note that
GC is C 1 and its domain is open. Further, note that C = G−1
C ({0}).

• By theorem 4.6.12, if the maximum of f on S occurs on C then it must satisfy the 2 constraint
Lagrange system. The Lagrange multiplier system for C is

3x2 − 2y = −2λx − 2x + 2y = −2λy 2=µ x2 + y 2 = 1 z=0

• Multiplying the first equation by y and the second by x gives 3x2 y − 2y 2 = −2x2 + 2xy. Since
x2 + y 2 = 1, this gives 3y − 3y 3 − 2y 2 = −2 + 2y 2 + 2xy, so x = 3 − 3y 2 − 4y + y2 . Substituting
this into x2 + y 2 = 1, WolframAlpha gives us that this has no solutions.

Comparing points:

• From our calculations above, the maximum of f must occur at one of (0, 0, 0), (0, 0, 1), 32 , 23 , 0 ,

and 32 , − 23 , 19 .

• We compute f (0, 0, 0) = 0, f (0, 0, 1) = 2, f 23 , 23 , 0 = − 27

4
, and f 23 , − 32 , 19 = 50



27 . Therefore
f has a maximum of 2 on S at (0, 0, 1).

4. For a fixed A ∈ R, define the curve CA in R2 by the equation

y 2 − 2xy = x4 + Ax + 4.

You can view this family of curves on Desmos. Note this demo will not help you justify your answers
below.
(a) Prove that if A ̸= −6 and A ̸= 6, then CA is a regular curve (smooth manifold of dimension 1).
You may use WolframAlpha to solve a 1-variable quartic equation.

Solution:

• Let A ∈ R \ {−6, 6} and define F : R2 → R by F (x, y) = y 2 − 2xy − x4 − Ax − 4. Note

that F is a polynomial and hence is C 1 on R2 .
• Furthermore, note that CA = {(x, y) ∈ R2 : y 2 − 2x = x4 + Ax + 4} = F −1 ({0}) and that
CA is non-empty since (0, 2) ∈ CA .

• We compute the Jacobian of F as

DF (x, y) = −2y − 4x3 − A 2y − 2x

 

• Suppose there exists (x, y) ∈ CA such that DF (x, y) has no invertible submatrices. Then
−2y − 4x3 − A = 0 and 2y − 2x = 0. This gives y = x and A = −2x − 4x3 . Substituting
this into the original equation gives

x2 − 2x2 = x4 + (−2x − 4x3 )x + 4.

• From WolframAlpha, we see that the solutions to this quartic equation are x = 1 and
x = −1. Since A = −2x − 4x3 , we then have A = 6 and A = −6.
• Therefore, if there exist (x, y) ∈ CA such that DF (x, y) = (0 0), then A = 6 or A = −6.

Page 5
 • Since we have assumed that A ̸= −6, 6, we have that DF (x, y) ̸= (0 0) for all (x, y, z) ∈
CA .
• Therefore, for all (x, y, z) ∈ CA , RankDF (x, y) = 1.

• Therefore CA is a regular curve by the implicit function theorem (or by theorem 4.5.7 if
you prefer).

(b) This family of curves can instead be viewed as the z-slices of the surface S in R3 defined by the
equation

y 2 − 2xy = x4 + xz + 4
Show that this equation defines a 2-dimensional smooth manifold.

Solution:

• Define G : R3 → R by G(x, y, z) = y 2 − 2xy − x4 − xz − 4. Note that G is a polynomial

and hence is C 1 on R3 .
• Also, note that S = {(x, y, z) ∈ R3 : y 2 − 2xy = x4 + xz + 4} = G−1 ({0}). Furthermore,
note that S is non-empty since (0, 2, 0) ∈ S.
• We compute the Jacobian of G as

DG(x, y, z) = (−2y + 4x3 + z 2y − 2x x)

• Suppose (x, y, z) ∈ S so that DG(x, y, z) is not full rank. Then −2y + 4x3 + z = 2y − 2x =
x = 0. Solving this system gives x = y = z = 0. However, note that (0, 0, 0) ̸∈ S.

• Therefore, for all (x, y, z) ∈ S, we have that DG(x, y, z) is full rank.

• Therefore S is a regular surface of dimension 2 by the implicit function theorem (or by
theorem 4.5.7 if you prefer).

(c) The curves C6 and C−6 are regular at each point except (−1, −1) and (1, 1) respectively. These
points where C6 and C−6 fail to be regular are called singularities. View this surface and family
of curves on Math3d. (Google Chrome is the most stable browser for Math3D.) Informally explain
how the shape of the surface appears to relate to the singularities of C6 and C−6 . Your explanation
must discuss both the surface and the curves.

Solution:

• For z > 6 and z < −6, then z-slice has two pieces where y is a C 1 function of x, whereas
for z ∈ (−6, 6) the z-slice has two pieces where x is a C 1 function of y.
• For z = −6 and z = −6, the slice is path-connected but can be broken into two pieces
where x is a function of y or two pieces where y is a function of x. Note that these
functions are not C 1 .
• At z = 6 and z = −6, the two pieces join at ”saddle points” of the surface: locally z is a
function of x, y at the points (1, 1, −6) and (−1, −1, 6) and those points are saddle points
of those functions. I.e., the saddle points correspond to singularities of the curves CA .
• Intuitively, the surface is joining together at those two points to form a handle.

Page 6